P
Peng Dai
Researcher at South China University of Technology
Publications - 55
Citations - 1516
Peng Dai is an academic researcher from South China University of Technology. The author has contributed to research in topics: Graph (abstract data type) & Markov decision process. The author has an hindex of 14, co-authored 53 publications receiving 930 citations. Previous affiliations of Peng Dai include University of Washington & Google.
Papers
More filters
Proceedings Article
Decision-theoretic control of crowd-sourced workflows
Peng Dai,Daniel S. Weld +1 more
TL;DR: A planner is described, TURKONTROL, that formulates workflow control as a decision-theoretic optimization problem, trading off the implicit quality of a solution artifact against the cost for workers to achieve it.
Journal ArticleDOI
POMDP-based control of workflows for crowdsourcing
TL;DR: In this paper, decision-theoretic techniques for the problem of optimizing workflows used in crowdsourcing are presented, where AI agents that use Bayesian network learning and inference in combination with Partially-Observable Markov Decision Processes (POMDPs) for obtaining excellent cost-quality tradeoffs.
Proceedings Article
Artificial intelligence for artificial artificial intelligence
Peng Dai,Daniel S. Weld +1 more
TL;DR: An end-to-end system that first learns TURKONTROL's POMDP parameters from real Mechanical Turk data, and then applies the model to dynamically optimize live tasks, which produces significantly superior artifacts compared to those generated through nonadaptive workflows using the same amount of money.
Proceedings ArticleDOI
And Now for Something Completely Different: Improving Crowdsourcing Workflows with Micro-Diversions
TL;DR: This work proposes an investigation into how to use diversions containing small amounts of entertainment to improve crowd workers' experiences and finds that micro-diversions can significantly improve worker retention rate while retaining the same work quality.
Proceedings ArticleDOI
Graph Meta Network for Multi-Behavior Recommendation.
TL;DR: In this article, a multi-behavior recommendation framework with graph meta network is proposed to incorporate the multi-behavior pattern modeling into a meta-learning paradigm, which automatically distills the behavior heterogeneity and interaction diversity for recommendations.